C. C. Mundt - plantpath.cornell.eduplantpath.cornell.edu/Courses/pp601/MGM lectures 2004/Mundt 2002... · 5 Jul 2002 10:25 AR AR165-PY40-14.tex AR165-PY40-14.SGM LaTeX2e(2002/01/18)

5 Jul 2002 10:25 AR AR165-PY40-14.tex AR165-PY40-14.SGM LaTeX2e(2002/01/18)P1: GJB10.1146/annurev.phyto.40.011402.113723

Annu. Rev. Phytopathol. 2002. 40:381–410doi: 10.1146/annurev.phyto.40.011402.113723

Copyright c© 2002 by Annual Reviews. All rights reserved

USE OF MULTILINE CULTIVARS AND CULTIVAR

MIXTURES FOR DISEASE MANAGEMENT

C. C. MundtDepartment of Botany and Plant Pathology, Oregon State University, Corvallis, Oregon97331-2902; e-mail: [email protected]

Key Words epidemiology, genetic diversity, host-pathogen interactions, pathogenevolution, sustainable agriculture

■ Abstract The usefulness of mixtures (multiline cultivars and cultivar mixtures)for disease management has been well demonstrated for rusts and powdery mildewsof small grain crops. Such mixtures are more useful under some epidemiologicalconditions than under others, and experimental methodology, especially problems ofscale, may be crucial in evaluating the potential efficacy of mixtures on disease. Thereare now examples of mixtures providing both low and high degrees of disease controlfor a wide range of pathosystems, including crops with large plants, and pathogensthat demonstrate low host specificity, or are splash dispersed, soilborne, or insectvectored. Though most analyses of pathogen evolution in mixtures consider static costsof virulence to be the main mechanism countering selection for pathogen complexity,many other potential mechanisms need to be investigated. Agronomic and marketingconsiderations must be carefully evaluated when implementing mixture approachesto crop management. Practical difficulties associated with mixtures have often beenoverestimated, however, and mixtures will likely play an increasingly important roleas we develop more sustainable agricultural systems.

INTRODUCTION

One of several crop diversification strategies for disease control (39, 40) is togrow mixtures of plants that differ in their reaction to a pathogen. The philoso-phy and concepts underlying the use of multiline cultivars (mixtures of lines bredfor phenotypic uniformity of agronomic traits) and cultivar mixtures (mixtures ofagronomically compatible cultivars with no additional breeding for phenotypicuniformity) were described elegantly in previousAnnual Review of Phytopathol-ogychapters (17, 167). Recent minireviews have summarized the epidemiologicalimpacts of host diversity (45) and current research on cereal cultivar and speciesmixtures in Europe (36). Smithson & Lenne (153) have provided a thorough up-date of the literature on cultivar mixture effects on disease levels, yield, and yieldstability for a range of crops. A website useful for teaching about mixtures inadvanced plant pathology classes is now available (23).

0066-4286/02/0901-0381$14.00 381

5 Jul 2002 10:25 AR AR165-PY40-14.tex AR165-PY40-14.SGM LaTeX2e(2002/01/18)P1: GJB

382 MUNDT

A critical point that has previously been emphasized is that of functional diver-sity, i.e., disease reductions will not be obtained with all mixtures, and randomlychosen mixture components will not necessarily provide adequate disease control.Rather, mixture components need to be relevant, or functional, to the pathogenpopulation in question (112, 147, 167). The effects of mixtures on disease inten-sity as compared to their component pure stands can range from a disease in-crease to nearly complete disease control (153). In this review, I discuss potentialexplanations for this large variation in mixture effects, both within and betweenpathosystems. I then discuss the effects of host diversity on pathogen evolution, andcomment on considerations pertinent to implementation of diversity approachesin agricultural production. The term mixture is used to include both multilinecultivars and cultivar mixtures. Though intercropping (mixtures of different cropspecies) is important (14, 44), especially in traditional agriculture (157), that largeand complex literature is not discussed here.

SPECIALIZED, FOLIAR PATHOGENS OF SMALL GRAINS

Most studies regarding the effects of host mixtures on disease have been withspecialized, polycyclic, foliar pathogens of small grains, primarily rusts (Puc-cinia spp.) and powdery mildews (Blumeria [=Erysiphe] graminis), but therehave been recent studies with rice (Oryzae sativa) blast, caused byMagnaporthegrisea(72, 103). These extensively grown, self-pollinated cereal crops have a longhistory of “boom-and-bust” cycles of effective control of disease followed by fail-ure, as pathogens adapt to widely-deployed single race-specific resistance genes(13, 17, 68, 173). Host resistance is often the only economically viable controlmethod for these diseases. Thus, strategies to increase resistance durability againstthese diseases have been of particular interest.

Several mechanisms have been postulated to explain the reduction in severityof disease caused by polycyclic, foliar pathogens that interact in a gene-for-genemanner with their hosts when mixtures are grown (17, 27, 45, 72, 112, 167). Thedilution of inoculum that occurs due to increased distance between plants of thesame genotype often appears to be the most important mechanism (20, 27, 167).However, other mechanisms may also be important (27, 72). For example, inducedresistance, which is predicted to be important by simulation models (79), accountedfor about 30% of the total reduction of yellow rust (caused byPuccinia striiformis)in wheat (Triticum aestivum) cultivar mixtures in the field (22). In another studywith wheat yellow rust, cultivars resistant to all inoculated races compensatedfor susceptible cultivars through increased tiller number, and this compensationsometimes accounted for a very large percentage of the observed reduction indisease severity (37, 38). When each of the cultivars was susceptible to one ormore races, however, this same effect was not seen (2). By comparing barley(Hordeum vulgare) cultivar mixtures containing either the same or different race-specific resistance genes, Wolfe et al. (170) estimated that differences in geneticbackground among cultivars contributed an additional 33% reduction of powdery


CULTIVAR MIXTURES 383

mildew beyond that provided by diversity for race-specific resistance. Thoughmuch is known about the efficacy of mixtures against foliar small grain diseases,there is still substantial variation in disease control attained in different studies. Forexample, powdery mildew of small grains is an ideal target for use of host mixtures,owing to a relative abundance of single-gene resistances, a small plant size, shallowdispersal gradient of the pathogen, and a large number of pathogen generationsper crop season (45). Even in this case, however, there can be substantial variationin disease control among mixture studies. For example, Smithson & Lenne (153)noted disease reductions relative to pure stands ranging from 4% to 89%. Reasonsfor such large variation in the efficacy of mixtures on disease are discussed below.

Mixture Composition

For diversity to be functional, there must be an appropriate match between theresistance genes incorporated in a mixture and the avirulence genes present inthe target pathogen population. Thus, a matrix of host and pathogen genotypes isusually considered, in an attempt to minimize the percentage of the host populationtoward which a given race will be virulent. Such information has been used toconstruct, and alter over time, the resistance genes utilized in multiline cultivars(18, 72). In the United Kingdom, tables have been provided to farmers in whichsmall grain cultivars are placed into different diversification groups, based on theirreactions to current virulence combinations ofP. striiformisandB. graminisf. sp.hordei(142, 143). A similar process is almost always used in choosing resistancegenotypes to include in mixtures, even if not in a formal way. Statistical approachesfor choosing mixture components based on field analysis of two-way combinationsof cultivars (71, 86) and of the local pathogen population (176) have been proposed.Molecular analyses of pathogen populations (177) and individual avirulence genes(82, 162) may provide further insights into the most appropriate resistance genesto deploy in host mixtures to maximize durability.

Given pathogen specificity, disease severity will decrease with decreasing fre-quency of a host genotype in mixture, as has been demonstrated numerous timeswith rusts and powdery mildews of small grains (17, 112, 167), and with rice blast(72, 103). Leonard’s models (83) predict that the apparent infection rate (161) willdecline logarithmically with the proportion of a genotype in mixture. This loga-rithmic relationship has been confirmed in the field for several small grain diseases(20, 33, 73, 83, 87, 88). Both logarithmic (36) and linear (2, 137) relationships havebeen reported between disease severity and host frequency.

Many experimental studies of mixtures have involved a single pathogen raceand two host genotypes, with one host genotype being susceptible and the othercompletely resistant to that race. Such studies are useful to investigate the dilutioneffect of mixtures on disease, and are sometimes of practical use to protect anagronomically desirable but susceptible genotype by mixing it with a resistant, butless desirable, genotype (45). Most practical applications, however, involve a morecomplex matrix of host and pathogen genotypes, wherein each host genotype issusceptible to one or more pathogen races and each race may be virulent to more


384 MUNDT

than one host genotype. This allows for the use of resistance genes that have been“defeated,” and for mechanisms such as induced resistance and interracial compe-tition to function (112). In these differentially susceptible (45) mixtures, one wouldexpect disease to decline with number of mixture components, if these componentsreact differently to the pathogen genotypes in question. Though studied little, sucha decline has been demonstrated in two cases (103, 122). However, these wereaverage effects, and there is sufficient variation among specific mixtures such thata comparison of any two mixtures with differing number of genotypes may notdemonstrate the expected result.

Given the logarithmic relationship between infection rate and host frequencydescribed above, one would expect diminishing returns to decreasing frequency ofa genotype in mixture or to an increase in the number of genotypes used in a mixture(83, 101). On the other hand, including a large number of genotypes in a mixturecould be beneficial in slowing pathogen evolution toward complex virulence (91).In reality, practical considerations often play a large role. Genotypes included ina cultivar mixture usually must be agronomically competitive on their own, andmixing is usually done by farmers themselves or by seedsmen. Thus, the numberof genotypes included in a cultivar mixture is usually very limited, often betweentwo and four. The agronomic performance of genotypes included in a multilinecultivar is usually quite similar. In addition, individual lines are usually maintainedby some type of certifying organization and provided to seed growers as a bulkpopulation. As a consequence, as many as 8 to 12 lines have been included in somemultiline cultivars (4, 18).

The slope of the relationship between a disease parameter (infection rate,% severity, etc.) and host frequency determines the strength of the mixture’s effecton disease and will depend on several factors, including those discussed below.

Epidemic Intensity

Among the more important outcomes of Leonard’s (83) model is that diseasereduction due to mixing will increase with increasing generations of pathogenreproduction. For example, in an equal mixture of susceptible and immune plants,the predicted disease level on susceptible plants in the mixture relative to a purestand of susceptible plants will be 0.5 in the first generation of secondary spread,0.25 in the second generation, 0.125 in the third generation, etc. The number ofpathogen generations that occur in a given epidemic will depend on the level ofinitial infection, the generation time of the pathogen, and the rate of epidemicdevelopment. The usual pattern seen in the field is for disease levels to initiallydiverge over time, as predicted by Leonard’s (83) equations, and then to converge asthe host’s carrying capacity for disease is approached [for example, see (169)]. Thereason for the convergence is that the absolute rate of disease increase eventuallybecomes lower in the pure stands than in the mixture as availability of uninfectedhost tissue in the more heavily infected pure stands limits epidemic progression(101).



Epidemic speed is also crucial. The more rapidly the host’s carrying capacityfor disease is approached, the less effective a mixture will be (101). For example,Alexander et al. (3) did not observe the level of stem rust (caused byPucciniagraminis) control they expected in wheat cultivar mixtures. Stem rust epidemicsusually begin late in the growing season, the pathogen has a relatively long genera-tion time, and rates of stem rust epidemics are often high. Based on Leonard’s (83)exponential model, a disease reduction of 75% was expected (104) in Alexanderet al.’s (3) plots, which consisted of 40% susceptible and 60% resistant plants.When logistic growth was assumed to account for the effects of limited carryingcapacity, however, the predicted disease reduction was only 26% (104), a valuevery close to that observed in the field (3). The impact of carrying capacity maybe even more extreme if one accounts for the important effect of latent infectionson available infection sites (43). On the other hand, the rate of approach to carry-ing capacity may be much less significant if disease increase is discontinuous, asopposed to the continuous progression assumed by most analytical models of epi-demic progression (45). Thus, divergence and subsequent convergence of diseaselevels in mixtures versus pure stands is likely to be the typical pattern seen in thefield in a qualitative sense, but will differ quantitatively among pathosystems andenvironments.

As the major effect of mixtures is to reduce the infection efficiency of thepathogen through the dilution effect (see above), large amounts of external inocu-lum can greatly reduce the efficacy of a mixture for disease control. For exam-ple, year-round potato production in one region of Ecuador apparently resultedin large amounts of outside inoculum that masked the effect of potato (Solanumtuberosum) cultivar mixtures on late blight, caused byPhytophthora infestans. Incontrast, mixture effects on late blight were greater at a site more distant from com-mercial potato production in Ecuador (48), and in more temperate areas of Peru(K.A. Garrett, L.N. Zuniga, E. Roncal, G.A. Forbes, C.C. Mundt & R.J. Nelson,in preparation) and the United States (47). A given level of outside inoculum willconstitute a larger proportion of total inoculum in a mild epidemic than in a severeone (45), but the opposite could be true when conditions are favorable for epidemicdevelopment, and hence inoculum production, at a regional level.

The influence of epidemic intensity on mixture efficacy will thus depend on theunderlying cause(s) of disease increase. If high disease incidence in pure stands isdue to a fast approach to carrying capacity or large amounts of outside inoculum(either initially or on a continuing basis), then mixtures may be less effective insevere than in less severe epidemics. If, on the other hand, high disease incidence inpure stands is being driven by the number of pathogen generations, then mixturesmay be more effective in severe than in less severe epidemics.

Spatial Scale

Disease control attained through host diversification can be influenced greatlyby spatial scale of deployment, and it has been suggested that mixture efficacy


386 MUNDT

is greater in production-scale situations than in small-scale experimental plots(103, 105, 167). There are several potential reasons for this spatial influence, threeof which are discussed below.

EXPERIMENTAL PROCEDURE AND DESIGN Experimental plots are often artificiallyinoculated, and at much higher concentrations than would be normal in nature, in anattempt to assure disease establishment and to avoid stochastic effects in diseaseestablishment. This practice can reduce the number of generations of pathogenincrease that occur before the crop’s carrying capacity for disease is reached, thusreducing the effectiveness of mixtures on disease. A similar situation may arisewhen infected “spreader plants” are placed within or around experimental plots(9, 36), with the additional disadvantage of ongoing inoculum production from thespreader plants throughout the epidemic.

Nearly all field experiments incorporate some type of “untreated” control, whichwill have a greater severity of disease and produce greater amounts of inoculumthan other plots. In the case of mixtures, untreated controls are usually pure standsof susceptible host genotypes. All types of disease control are affected by interplotinterference, including quantitative resistance (134) and fungicides (57). However,mixture experiments are especially vulnerable to interplot interference because theprimary effect of mixtures, which is to reduce the pathogen’s infection efficiencythrough dilution of inoculum, can be overwhelmed by large amounts of outsideinoculum, as discussed above. In contrast, many fungicides and quantitative re-sistance also impact the latent period, lesion expansion, and sporulation, whichare not directly affected by high concentrations of outside inoculation. The impor-tance of interplot interference depends on factors such as the ratio of plot size toarea between plots, wind conditions, etc. (103, 135, 167). The mixture effect cansometimes be obliterated by interplot interference (103, 167).

NATURE OF DISEASE SPREAD Minogue & Fry (96) and van den Bosch et al. (159,160) developed mathematical models that describe the movement of disease from afocus as a traveling wave of constant velocity. Later, Ferrandino (34) suggested thatepidemic velocity will increase in time and space for wind-dispersed pathogens.If disease spreads as a wave of constant velocity, then the difference in velocitybetween a pure stand and a mixture will not be affected by the spatial scale of study.If velocity increases in time and space, however, then the difference in velocitybetween a pure stand and a mixture will become greater as disease expands in timeand space (105). Consequently, disease control will be relatively more effectivewhen studied over larger spatial scales. Through consideration of previous theo-retical models (159, 160) and field studies with wheat yellow rust, van den Boschet al. (158) concluded, “In an ideal mixture of susceptible and resistant plants,the radial velocity of focus expansion increases linearly with the logarithm of theproportion of susceptible plants.” This logarithmic relationship was confirmed ina second study with wheat stripe rust (19) and a subsequent study with bean rust,caused byUromyces appendiculatus(5). It is not clear, however, whether velocities



were constant in these studies, as focus expansion was followed only until the ra-dius reached 2 m or less. In contrast, recent studies with wheat stripe rust overgreater distances (33 m) showed that epidemic velocity increases in time and space,and that this rate of change is greater for a pure stand than for a mixture (C.C.Mundt & L. Wallace, unpublished).

Many epidemics do not have a distinctly focal nature, and some may begin fromnumerous, randomly dispersed infections. Spatial scale may be of importance to theepidemiological results of diversity in these cases as well. It was earlier mentionedthat mixtures are highly sensitive to the effect of external inoculum. Thus, mixtureswith potentially functional diversity may not be as effective as desired in locationswith substantial external inoculum. If, on the other hand, deployment of mixturesreduces inoculum densities on a regional basis, there may be a positive feedback,since less external inoculum may be experienced by any given field as mixturesare grown more extensively in the region (113).

OBSERVATION OF MIXTURES AT LARGE SPATIAL SCALE Comparisons of mixtureperformance at different spatial scales have usually been restricted to casual ob-servations that mixtures provide better disease control at commercial scale thanin small experimental plots (103, 167, 169) or that yield loss in mixtures has notbeen reported under extensive commercial use (18). Large-scale experiments withreplicated treatments and appropriate controls are needed, but are prohibitivelyexpensive and/or logistically impossible in most cases. There are, however, twoexamples in which careful evaluation of the commercial implementation of mix-tures suggests that spatial scale may be of importance.

In the early 1980s, barley cultivar mixtures began to be used to control powderymildew in the then German Democratic Republic, where the incidence of barleyfields with severe mildew had reached about 50%. The percentage of fields sownto mixtures gradually increased to 92% (360,000 ha of mixtures) by 1990, whilethe incidence of severely infected fields declined to 10% and the percentage offields sprayed with fungicide was reduced about threefold. Similar declines inbarley powdery mildew were not observed over the same time period in adjacentcountries, where diversification was not practiced (168, 171).

In a second example, in China, mixtures of glutinous rice cultivars highly sus-ceptible to blast were mixed with non-glutinous cultivars that were more resistant,and compatible with a different set of blast races, than were the glutinous cultivars.Four mixtures were deployed in a total of 812 ha of contiguous rice fields spanningfive townships in 1998 and in 3342 ha of contiguous rice fields over ten townshipsin 1999. A replicate of pure-line controls was planted in each township. Thus,mixtures were deployed in large, contiguous areas, but monoculture controls oc-cupied a very small percentage of the total rice area, which minimized the effectsof interference caused by high disease densities in pure stands. Mixtures reducedthe severity of blast on susceptible glutinous rice cultivars by an average of 94%as compared to pure stands of the same cultivars (178). The practice expanded to25,000 ha in 2000 and 100,000 ha in 2001 (Y. Zhu, personal communication).


388 MUNDT

Though results from the above two cases are encouraging, it is also important tokeep in mind the “file drawer problem,” i.e., positive results tend to become known,whereas negative results may be placed in a file drawer and never published (29b).

FACTORS DETERMINING MIXTURE IMPACTSON OTHER PATHOSYSTEMS

Autoinfection/Alloinfection Ratio

Mixtures are unlikely to have significant effects on disease unless there is sub-stantial exchange of inoculum among different host genotypes in the population.Thus, the effectiveness of mixtures for disease control is expected to decline withincreasing proportion of autoinfection (8, 51, 120). Autoinfections are those “inwhich the donor (infector) host individual is the same as the recipient (infected)host individual,” whereas alloinfections are those “in which the donor (or infec-tor) host is a different individual from the recipient (or infected) host individual”(144). The proportion of autoinfection is influenced by both pathogen and hostcharacteristics, as discussed below.

Dispersal

Steep inoculum dispersal gradients can result in high levels of autoinfection, andmathematical models have shown mixture efficacy to decline with increasing steep-ness of the pathogen’s dispersal gradient (41, 68, 118, 174). There is some field evi-dence to support this contention. For example, bacterial pathogens are often splashdispersed, a process that can result in very steep dispersal gradients (42, 90). Mix-ture impacts less than those commonly recorded for rusts and powdery mildewsof small grains have been observed for both bacterial blight of beans (Phaseolusvulgaris), caused byXanthomonas campestrispv. vesicatoria(75), and bacterialblight of rice, caused byXanthomonas oryzaepv. oryzae(1). Such observationsdo not isolate the effect of dispersal gradient steepness, however. For example, asmall number of pathogen generations and substantial lesion expansion (see be-low) may reduce mixture effectiveness on bacterial blight of rice (1). In diseasecaused by splash-dispersed fungi, mixtures have also shown relatively low diseasecontrol (24, 50, 63, 93, 109, 115, 128, 150), though there are exceptions (60, 122).Comparisons are again confounded, as these splash-dispersed pathogens also tendto demonstrate less host specificity than is commonly seen with rusts and powderymildews, and some of the results were from crops with plants larger than that ofsmall grains.

Lesion Expansion

Continued expansion of an infection can contribute substantially to autoinfection,and is also a major contributor to total epidemic development (11). Computersimulations suggest that extensive lesion expansion can greatly reduce the efficacy



of mixtures on disease (77, 78). Studies in which environmental conditions andpathogen generation time were held constant showed that disease reductions inwheat mixtures were about twice as large for leaf rust (which has small, determinatelesions) as for wheat yellow rust (which demonstrates substantial lesion expansion)(77). In the field, approximately equal reductions of yellow rust and leaf rustwere found in the same wheat cultivar mixtures (103). These mixtures were notdesigned for protection against leaf rust, however, and diversity may have beenmore functional against yellow rust than leaf rust, thus biasing the comparison.

Host Geometry

All else constant, autoinfection would be expected to increase with increasinggenotype unit area (GUA), i.e., the ground area occupied by an independent, gene-tically homogeneous unit of host tissue (111). In a random mixture of host geno-types, the genotype unit is a plant, and GUA is the ground area occupied by anindividual plant. Similarly, if host genotypes are instead deployed in alternatingrows or in different fields, then GUA is the area occupied by a row and field,respectively. GUA does not account for intermingling of adjacent plants or three-dimensional dispersal of inoculum. It can, however, provide a useful approximationof the degree of mixing, or the “grain” of diversity sensu Pielou (138). Computersimulations and field studies showed that mixture efficacy decreases when the de-gree of aggregation of plant genotypes is changed to alter GUA (111, 117–120,174). On average, rice mixtures controlled blast somewhat more effectively whenplants were mixed within, rather than between, hills (72, 103) or rows (72). Ran-dom mixtures of sorghum (Sorghum bicolor) cultivars provided better control ofColletotrichum sublineolumandExserohilum turcicumthan did alternating rows ofthe same two cultivars (129). Similar results were found with yellow rust of wheat(16). As noted above, row mixtures of different rice cultivars were highly effectiveat controlling blast in China (178). It was impractical to include a comparison withrandom mixtures in that study, however.

GUA interacts with other spatial variables to determine disease severity in mix-tures. Computer simulations (120) and field studies with rust pathogens (111, 117,119) showed that GUA had little influence on mixture efficacy if initial inoculumwas distributed in a single focus, rather than uniformly over plots. Later theoreticalwork suggested that the number of host units may be more important than theirsize, and that mixtures of large genotype units can provide substantial disease con-trol, even when inoculum is distributed uniformly, provided that the total numberof host units is sufficiently large (107). This is a difficult concept to test in the field,and results may be affected by shape and spatial pattern of genotype units, as wellas by environmental conditions such as wind speed (108).

Overall, there is substantial variation in reported levels of disease control bymixtures for crops with plants larger than small grain cereals (7, 24, 75, 89, 92, 97,119, 128, 129, 152, 156, 166). This is perhaps not surprising, given the large num-ber of host and pathogen variables among the diseases represented. Nonetheless,there is an indication that mixtures may sometimes be of great value for controlling


390 MUNDT

diseases of crops with large plants, even for tree species. Computer simulationssuggest that mixtures of apple (Malus× domestica) genotypes could reduce theseverity of scab (caused byVenturia inequalis) from 86% to 34% (12). Prelimi-nary field results showed 59% to 95% reductions of scab incidence on the cultivarGolden Delicious when grown in alternating rows with two other cultivars, thoughmixtures alone were not sufficient to provide commercially acceptable levels ofcontrol (89). In some cases, divergent results have been reported for highly similarpathosystems. For example, rust caused byMelampsoraspp. was reduced by 50%or greater in mixtures of willow (Salixspp.) clones (92). In contrast, mixtures ofpoplar (Populusspp.) clones reduced infections ofMelampsora larici-populinaby 10% or less throughout most of the growing season (97). Conflicting resultsbetween the twoMelampsorastudies may have resulted from differences in ef-fectiveness of the specific resistance genes deployed in each study or by differingexperimental designs and procedures. In the case of the willow rust study (92),mixtures consisted of 5–6 clones, minimum plot size was 0.5 ha, and inocula-tion was natural. For poplar rust, the mixture was of three clones, plot size wasonly 2.8× 3 m, and epidemics were initiated with high levels of inoculum (97).Multiline cultivars of coffee (Arabicaspp.) are used for protection against coffeerust (caused byHemileia vastatrix) on 350,000 ha in Colombia (36, 98). This isa proactive program to protect the crop before the disease arrives and, thus, theepidemiological effects cannot be evaluated at this time.

It has been hypothesized that mixture effects on disease will be greater at highplanting densities, owing to decreased plant size and, hence, decreased autoinfec-tion (10). The influence of plant density on disease levels in pure stands must alsobe accounted for, however. Though it often is expected that disease severity willincrease with plant density, field results are highly variable (14, 21). For example,the severity of barley powdery mildew declined with increased plant density, per-haps due to the decreased nutritional status of plants at high density (35). Thus,to isolate the effect of plant density on mixture efficacy for disease control, oneneeds to include all pure stands and all mixtures at a range of densities, a procedurethat is often difficult to accomplish in the field (35). To my knowledge, this hasbeen done in only one study. The effect of wheat cultivar mixtures on yellow rustseverity was greatest at an intermediate planting density in two years, despite thefact that rust severity in pure stands increased with planting density in one yearand decreased with planting density in the other year of the study (46).

Degree of Host-Pathogen Specificity

There are two situations in which the influence of host-pathogen specificity isrelevant to mixtures. The first is for mixtures of host genotypes with partial re-sistance against pathogens that can also express a high degree of specificity, e.g.,rusts and powdery mildews. The second situation is for host-pathogen systemsthat, as a whole, do not display the high degree of specificity typical of obligateparasites. For example,Rhynchosporium secalis, the causal agent of barley scald,can demonstrate host specificity (93), but this specificity is less stringent than for



biotrophic fungi and there are fewer sources of specific resistance available (122).In either situation, we are usually concerned with quantitative differences amonghost genotypes in mixture.

One approach to predict the epidemiological effect of such mixtures is to con-sider resistance to be qualitatively no different from specific resistance, and tosimply account for the difference in resistance level between mixture components(45). Other more mechanistic approaches (51, 52, 59) account for differences inresistance components (e.g., infection frequency, latent period, and sporulation)between host components. These models predict that mixtures of partially resistanthost genotypes can decrease, increase, or have no effect on disease severity rela-tive to their component pure stands, depending on the relative levels of resistancecomponents between host genotypes. Given that most pathosystems demonstratesome degree of specificity, quantitative adaptation of the pathogen to the differinghost genetic backgrounds in mixtures (26, 76, 114, 170) could also lead to diseasereductions in mixtures (26, 114).

Models that predict variability in performance among mixtures of host geno-types with partial resistance seem to better match field results. In the field, responsesranging from disease decreases to disease increases have been reported among mix-tures of partially resistant genotypes for barley powdery mildew (124), barley scald(93), and Septoria tritici blotch of wheat (29a, 109; C. Cowger & C.C. Mundt, inpreparation). In all of these studies, there were mixtures that showed disease in-creases in one year and disease decreases in another, perhaps owing to interactionsof resistance components with environment. Averaged over all mixtures investi-gated, the majority of studies of mixtures of partially resistant host genotypes haveshown relatively low levels of disease control (32, 50, 56, 63, 93, 109, 115, 124)or disease increases (125), though some have demonstrated reductions in dis-ease similar to that expected with race-specific resistance and obligate parasites(60, 122, 150).

Diseases Caused by Viruses

There have been very few studies regarding the effects of mixtures on diseasescaused by viruses. The outcome of mixtures on such diseases may be complicatedby effects on vector abundance and behavior, as influenced by interactions with thetransmission mode of the pathogen (140). Power (141) found that a 1:1 mixture ofsusceptible:resistant oat (Avena sativa) cultivars reduced the incidence of yellowdwarf to approximately the level observed in the resistant component for all threeyears of the study. In a different study of yellow dwarf, oat cultivar mixtures reduceddisease incidence below the mean of the component pure stands for populationsincluding two moderately resistant cultivars, but not in mixtures containing a highlysusceptible and a moderately resistant cultivar (64).

Reductions of yellow dwarf incidence in the oat cultivar mixtures in Power’s(141) study may have been caused by an interaction between vector behavior andvirus transmission. Aphids in cultivar mixtures showed higher movement rates andshorter tenure times than aphids in pure stands. This may have reduced transmission


392 MUNDT

rates for this virus, because an aphid must feed for several hours to inoculate a hostplant (141). The transmission dynamics of other pathogens could provide verydifferent results. For example, maize (Zea mays) grown in association with beansgreatly reduced the density of the leafhopper vector of the corn stunt spiroplasma,but not incidence of the disease. In this case, increased movement and decreasedtenure times may have raised the number of transmissions per leafhopper, as thepathogen can be transmitted after very short feeding times by its vector (139).

The incidence of wheat soilborne mosaic was reduced by 33.2 and 39.8%,respectively, in 1:1 and 1:3 mixtures of susceptible:resistant cultivars. Further,the ELISA index of infected plants was 27.1% and 33.7% less in the 1:1 and1:3 mixtures, respectively. Under field conditions, the resistant cultivar did notproduce viruliferous zoospores of the soilborne vector,Polymyxa graminis(54).Thus, presence of the resistant cultivar in the mixture may have reduced virustransmission between susceptible plants in the mixture.

Soilborne Pathogens

The impact of mixtures on soilborne pathogens is likely to be determined by theextent to which secondary cycles of the pathogen occur, degree of host speci-ficity, and spatial pattern of the pathogen in soil. As noted above, mixtures hada significant effect on wheat soilborne mosaic, a viral disease with a soilborne,protist vector (54). Mixtures of a resistant and a susceptible oat isoline decreasedthe disease index of Victoria blight and spore production ofHelminthosporiumvictoriae in both field and greenhouse soil infested in foci (6). Crown and rootrot of sugar beet (Beta vulgaris) caused byRhizoctonia solaniwas reduced bymixing a susceptible cultivar with a highly resistant, but lower-yielding cultivar.Mixtures containing between 1/6 to 1/3 of the resistant cultivar provided yieldssuperior to that of the susceptible cultivar in the presence of disease and yieldedcomparably to the susceptible cultivar in the absence of disease. This practice isnow being rapidly adopted by sugar beet growers in Michigan (53; J.M. Halloin,personal communication). There are three published reports from different locali-ties involving eyespot of wheat, caused byPseudocercosporella herpotrichoides.All three studies included one or more mixtures of a cultivar possessing a majorgene for resistance toP. herpotrichoidesmixed with a susceptible cultivar. Resultsranged from a statistically significant 13% reduction of eyespot severity (110) tosmaller effects that were nonsignificant (109, 110, 146). Cultivar mixtures did notsignificantly reduce severity of eyespot in barley (50).

Interactions between plant genotypes in mixtures can sometimes provide yieldbenefits in the presence of soilborne pathogens, even when disease levels are notaffected. The yield of mixtures of resistant and susceptible soybean (Glycine max)isolines at four of six locations was equal to that of the resistant pure stands inthe presence ofPhytophthora sojaebecause resistant isolines compensated foryield reductions as high as 35% experienced by susceptible isolines in mixture;substantial, though not complete, compensation occurred at the other two locations(165). Large yield increases over component means have often been recorded in



on-farm trials for wheat cultivar mixtures grown in the presence ofCephalosporiumgramineum, despite the lack of an effect of mixtures on severity of Cephalosporiumstripe disease (106, 116). This result was likely not due to simple compensation,however, as mixtures with the largest increases in yield were sometimes thoseconsisting of cultivars with similar levels of resistance to Cephalosporium stripe.Averaged over three seasons, one wheat cultivar mixture yielded 9.1% greater thanthe mean of its component pure stands when inoculated withP. herpotrichoides,despite no effect on eyespot severity; the mean yield increase of this mixture innoninoculated plots was only 2.9% (110).

PATHOGEN EVOLUTION IN MIXTURES

As discussed above, race-specific resistance genes can continue to provide largeepidemiological repercussions when deployed in mixtures, even after the matchingvirulence has evolved in the pathogen population. Will diversification also slowevolution to matching virulence of a given resistance gene? There are two issuesto be considered. The first and simpler one is whether a given resistance gene willbe more durable when deployed in a mixture than in a monoculture. A resistancegene deployed in a mixture will have less exposure to the pathogen populationthan if the same gene were deployed in monoculture of the same total crop area.This would be expected to reduce selection pressure and increase durability of thatgene; limited field observations seem to support this view (103).

A more difficult question is whether a given number of resistance genes will bemore durable if deployed in a mixture as compared to the same number of genesdeployed sequentially in monoculture or combined into a single host genotype.Not surprisingly, mixtures support more diverse pathogen populations than do purestands (18, 30, 72, 100), and the degree of diversity maintained within mixturesappears to be positively related to the degree of disease control provided (30). In thelonger term, the question is whether mixtures will select for increased frequency ofcomplex races, i.e., those with virulence corresponding to more than one resistancegene in the mixture. Unfortunately, this question is extremely difficult to testexperimentally, and may require large areas and/or large amounts of time to addressadequately.

In small-scale experimental plots, the relative frequency of complexB. grami-nis genotypes was greater in barley cultivar mixtures than in pure stand controls(26, 55, 121). Huang et al. (55) found that selection for complexity was less intensein row mixtures than in random mixtures of barley cultivars, supporting mathe-matical models (8, 51, 132) that predict less intense selection for complexity asautoinfection increases. At a larger spatial scale, the fungus increased in com-plexity in the former German Democratic Republic, where barley mixtures werebeing grown on as much as 360,000 ha. However, the same trend occurred in theformer Czechoslovakia, where mixtures were not grown (171). Thus, it has beensuggested that this increased complexity was an effect of migration, rather thanselection by the mixtures (172).


394 MUNDT

It is more difficult to determine the selective influence of mixtures for clonalpathogens because virulence genes will not be randomly associated with othergenes that influence pathogen fitness. In artificially inoculated plots, the proportionof complex races present in a mixture depended on the environment, host genotypesincluded in the mixture, and the genetic background of the complex race (30).In a very carefully designed greenhouse experiment, Kolmer (74) attempted todissociate genetic background from virulence by first passing a population ofPuccinia graministhrough the sexual stage. Populations of the pathogen werethen cycled on three different wheat multilines over 12 generations. He foundstrong selection for pathogen genotypes with virulence corresponding to three ofthe five resistance genes present in the multiline populations, but genotypes withfour or five corresponding virulences represented a small minority and occurredat frequencies lower than predicted by a mathematical model. Given that the samevirulences were selected on the susceptible, recurrent parent, this selection mayhave been due more to fitness effects associated with specific combinations ofvirulences than to any selective effect of the multilines. Thus, even in a well-designed experiment, it can be very difficult to make generalizations concerningthe effects of mixtures on evolution of clonal pathogens.

Chin & Wolfe (26) have demonstrated that the manner in which genotype fre-quency data are expressed can strongly influence interpretation of evolutionarytrends in mixtures. They found that the greater relative frequency of complexB. graminisgenotypes in barley cultivar mixtures was attributable to the mixturesbeing highly effective in reducing the absolute frequency (number of coloniesper tiller) of simple genotypes. In fact, the absolute frequency of the most com-plex genotype was less in a mixture of three barley cultivars than in the componentpure stands, perhaps owing to the selective influence of different host genetic back-grounds in the mixtures (see below). After several years of extensive commercialuse of barley cultivar mixtures in eastern Germany, there was an increase in the rel-ative, but not absolute, frequency ofB. graminisgenotypes combining virulence totwo of the three resistance genes most commonly used in the mixtures. Combiningdisease control and race frequency data in the wheat yellow rust study describedabove (30) also indicates a reduced absolute frequency of complex races in mixtureas compared to pure stand, though data from only one susceptible pure stand wasincluded in that study. It is thus possible that little loss of disease control wouldoccur, despite an increase in relative frequency of complex pathogen genotypesin mixtures. For example, inoculation of an oat multiline with a race ofPucciniacoronatavirulent against all six resistance genes did not diminish the ability of themixture to reduce disease severity (29 cited in 17). These results are all consistentwith observations of natural ecosystems, in which complex races are the most fre-quent but do not displace all other genotypes (31, 148) and are not sufficiently fit tocause high levels of disease (148, 149). Selection for increased fitness of complexraces could occur over multiple seasons in agricultural systems, if a mixture witha small number of host components were to be grown continuously (26).

The difficulty of studying pathogen evolution experimentally has encouragedthe use of mathematical models. Earlier modeling studies of pathogen evolution



in mixtures have been summarized (67, 84, 91), and there have been additionalcontributions subsequently (51, 52, 80, 81, 151, 154, 175). Most of these modelsincorporate the status quo view on this topic, which assumes that fitness costsassociated with virulence (more properly in most cases, a lack of avirulence) isthe only mechanism to counter selection for increased virulence complexity in thepathogen. The advantage to a complex race of being able to attack multiple hostgenotypes in a mixture is hypothesized to be countered by a reduction in fitnessassociated with lack of avirulence genes that correspond to the matching resistancegenes. In general, these models suggest that it will be difficult to prevent complexraces from eventually dominating the pathogen population, though the processmay be sufficiently slow such that the pathogen population can be managed.

Though models of pathogen evolution usually require a cost of virulence, evi-dence supporting such costs is very weak. Parlevliet (133) provided a thoroughreview of the literature regarding the cost-of-virulence concept, and found no asso-ciation of virulence with fitness as a general phenomenon. More recent molecularanalyses indicate that some avirulence genes can contribute to fitness pleiotropi-cally, and that inactivation of such avirulence genes can lead to reduced pathogenfitness. This does not appear to be the general case, however, and “a majorityof the known fungal avirulence genes have no obvious fitness function.” Further,redundancy of avirulence genes in plant pathogens may allow pathogens to avoidresistance gene recognition, while retaining fitness functions associated with avir-ulence (82).

Parlevliet (133) explained the general lack of association between virulenceand fitness in a two-step model consisting of mutation to virulence, which is oftenassociated with a fitness reduction initially, followed by selection for fitness modi-fiers that eventually ameliorate the initial fitness cost. He supported this model withempirical observations from several pathosystems. Though largely ignored in plantpathology, such fitness modifiers play a critical role in evolutionary processes ofother organisms. For example, selection for modifying genes can eliminate fitnessreductions initially associated with antibiotic resistance in bacteria (28, 85, 99).Thus, selection for virulence in mixtures is likely to be more complex than is ac-counted for by static fitness costs, and selection for fitness modifiers is likely tobe slower than selection for the virulence genes themselves (169). Further, it maybe important to estimate initial fitness costs associated with virulence before therehas been substantial selection for fitness modifiers.

Equally important to evaluating potential virulence costs is a need to examineother mechanisms (81, 84, 102, 112) that may counter selection towards increasedpathogen complexity in mixtures. A few examples of such mechanisms are dis-cussed below.

Alternative Mechanisms

Quantitative adaptation of pathogens to differing host genetic backgrounds presentin crop mixtures could play a very important role in suppressing selection for com-plex races. For example,B. graminisisolates able to attack two different major


396 MUNDT

resistance genes were collected from pure stands of barley and barley cultivar mix-tures containing these two resistances. Bulk populations of these isolates were thentested for infection efficiency on these same two host genotypes in the greenhouse.Populations derived from pure stands of the cultivar Hassan had a higher infec-tion efficiency on Hassan than on the cultivar Wing, and vice versa. Populationsderived from a mixture containing both Hassan and Wing had reduced infectionefficiency on both cultivars, suggesting that a form of disruptive selection pre-vented this complex race from obtaining high fitness on both host components inthe mixture (26). A similar result has recently been reported forBlumeria graminisf. sp.tritici (163) andMycosphaerella graminicola(114) on wheat, though subse-quent research did not demonstrate a consistent effect forM. graminicola(29a).Modeling studies suggest that differential adaptation to mixture components cangreatly slow selection for a complex pathotype (81), and that the rate of selectionfor complexity is slowed by increased genetic diversity for fitness of the pathogenand by increasing rates of autoinfection (76).

It is usually assumed that the relative fitness of competing pathogen genotypeswill be constant with regard to both population density and genotype frequency.However, differences in fitness components and effects of competition on thesecomponents can result in relative fitness being dependent on density and/or fre-quency (126, 127). In fact, modeling results indicate that relative fitness of com-peting pathogen genotypes will rarely be constant in nature (127). As pathogendensity and genotype frequency change during the course of an epidemic, relativefitness may change over time, perhaps even reversing the fitness ranking of sim-ple versus complex genotypes. Though experimental data are few, examples existof both density-dependent (62, 65, 126, 130) and frequency-dependent (65, 126)selection for competing genotypes of plant pathogens. Simulation modeling sug-gests that density-dependent effects can slow selection for complex races in hostmixtures, though density-dependence did not prevent the complex genotype fromeventually dominating the simulated pathogen populations (80, 81).

Plant pathogen populations commonly undergo severe genetic bottlenecks, e.g.,during overseasoning, and pathogen genotypes at low frequency may easily be lostowing to random drift during such times. Kiyosawa (67) suggested that genetic driftduring overseasoning could be critically important to the rate of “breakdown” ofresistant cultivars, and modeling results suggest that fungicide-resistant mutantscan be lost through such random effects (95). Pathogen genotypes combiningcomplex virulence with high fitness are likely to be in very low proportion initially.Thus, selection of highly fit, complex genotypes in mixtures could be greatlyslowed through stochastic effects when population size is small. To my knowledge,all models of pathogen evolution in mixtures are deterministic, and thus do notaccount for the effect of genetic bottlenecks.

Selection for complex pathogen genotypes will be reduced if a large proportionof the initial inoculum consists of migrants from another area where mixtures arenot grown. For example, immigration has a very large impact on the populationgenetics ofB. graminison small grains in Europe (171), and may have contributed



to the lack of strong selection for complex genotypes in barley cultivar mixturesgrown in the former German Democratic Republic (172). The importance of mi-gration will decline if similar mixtures are widely used, as a larger proportion ofimmigrants will have been exposed to the same selective pressure.

Selection in heterogeneous environments, a result of genotype× environmentinteraction, may slow selection for complex pathogen genotypes in mixtures. Forexample, race frequencies ofP. striiformiswere monitored in five three-way wheatcultivar mixtures planted at two locations and in two years (30). Complex racesable to attack two of three mixture components often dominated in one locationor year but not another. In no case did a complex race dominate in both years andlocations in which the study was conducted. For powdery mildew of barley, onthe other hand, the selective effect of cultivar mixtures on pathogen complexityshowed substantial consistency between seasons (26, 55), perhaps because sex-ual reproduction distributed virulence genes randomly among multiple geneticbackgrounds of the pathogen.

There will unquestionably be some selection toward increased relative fre-quency of complex races in host mixtures. After several decades of study, however,we still do not know the rate at which such selection will occur or to what degree, ifany, it would decrease the disease control provided by mixtures. Answers shouldemerge from increased field experimentation and increased commercial use ofmixtures. From a mechanistic standpoint, we need to abandon the old view ofstatic costs of virulence being the primary mechanism to counter selection forincreased pathogen complexity in mixtures. Other processes such as selection forfitness modifiers, disruptive selection, random drift, density- and/or frequency-dependent selection, genotype× environment interaction, or some combination ofthese factors will likely need to be accounted for.

Pathogen evolution can be managed to some extent by ensuring that no singlemixture is grown exclusively in either time or space (169). Though such manage-ment can be done purposefully [for example, see (18)], it will occur as a matterof course in many cases. For example, new genotypes can easily be added to cul-tivar mixtures as agronomically superior cultivars are released. In Oregon, wheatgrowers purchase or mix seed of new wheat mixtures containing recently releasedcultivars; some growers simply blend a new cultivar into the mixture that theyhad previously been growing (103; C.C. Mundt, unpublished). Spatial diversity ofmixtures is also likely to occur naturally, as different mixtures will perform betterin different localities within a region, owing to differences in soil, climate, andmanagement practices. In the former German Democratic Republic, eight barleycultivar mixtures incorporating 19 cultivars were utilized (171). In Washingtonstate, 18% of the common soft white winter wheat area was sown to cultivarmixtures in autumn 2000 (total of 106,000 ha of mixtures), and this included atleast 16 different mixtures (164). Even for perennial plantings, such as orchards,it may be possible to utilize different mixtures in time and space, since new plant-ings are continually being established over time, sometimes even within the sameorchard.


398 MUNDT

IMPLEMENTATION

Mixtures are being used increasingly in commercial production around the world(36). Many of the initial studies and applications of mixtures in “modern” agricul-ture were multiline cultivars (112). Multiline cultivars are still being used today,and may be especially useful when a strongly favored agronomic parent is availableor when crop uniformity is deemed crucial [for example, see (72, 98)]. In general,however, there has been a move in favor of cultivar mixtures over multilines. Cul-tivar mixtures are advantageous because they require no additional breeding effortand allow incorporation of agronomically superior cultivars as they become avail-able. Further, increased genetic diversity among components of a cultivar mixturemay provide additional control against the target disease, as well as some pro-tection from nontarget diseases and abiotic stresses (169, 170). Though much ofthe work on mixtures has involved self-pollinated crops, recent work with pearlmillet (Pennisetum glaucum) showed rust reductions and yield increases for mix-tures of inbred lines, and especially for random-mated populations and mixturesof two- and three-way crosses (156). Wilson et al. (166) recently demonstrateda “dynamic multiline” approach in pearl millet that combines the advantages ofboth multilines and gene pyramids. Selection over four cycles of open pollinationresulted in hybrid populations that reduced rust severity by 12–13% per cycle, andincreased digestible biomass yield by 4.1% per generation. Such an approach alsoallows for selection of plant genotypes that have coevolved in a mixed population,which may contribute to improved performance of mixtures. In fact, argumentshave been made in favor of selecting plant genotypes for their ability to performwell in mixtures for both self- and cross-pollinated species (39). Further researchis needed to determine the costs and benefits in taking such an approach.

A useful mixture must provide yield benefits as well as disease control. Yieldincreases of 1–5% are often provided by cultivar mixtures in the absence of sub-stantial disease, with larger increases when disease is of significance [for examples,see (38, 153, 169)]. Yield benefits can sometimes be substantially greater in largethan in small plots (49, 169; C.C. Mundt, unpublished), often for unknown rea-sons beyond that of disease control attained in plots of different size (49; C.C.Mundt, unpublished). Interactions between plant genotypes in mixture cause therelationship between disease level and yield of a mixture to be highly complex andunpredictable (36). In contrast, yield stability seems to be strongly and more con-sistently associated with mixtures (15, 66, 106, 136, 153, 167, among others). Yieldstability is often considered crucial in subsistence agriculture, and appropriatelyso. However, yield stability has also become crucial in industrialized agriculture,owing to the precarious economic situation facing commercial agriculture today.In fact, large, commercial growers in the United States are sometimes interested inmixtures more for their effect on yield stability than for yield level (66; C.C. Mundt,unpublished). Yield stability of mixtures could be due to disease control or com-pensation between host genotypes for damage caused by abiotic stresses such ascold injury (15, 36, 167), or other unknown genotype× environment interactions



(167, 169). Regardless of mechanism, it is usually impossible to predict at the timeof planting which host genotype will yield best in a particular season. Given thatsome yield advantage is expected to growing mixtures, on average, a farmer will in-crease the chances of attaining high yield by sowing a mixture of high-performingcultivars (167, 169). It seems clear that a competitive mixture must usually consistof genotypes with high yield potential in pure stand (103, 136, 169).

Though more effort is required to produce seed of mixtures than is the casefor single genotypes, there are several ways to provide mixed seed to farmers(103, 167). Multiline cultivars, which usually consist of several genotypes, aregenerally produced by a certifying agency, with lines being produced separately,bulked, and then provided to seed producers for increase and sale [see (4, 18)].Seed of cultivar mixtures, which usually consist of a smaller number of genotypes,is usually mixed by commercial seedsmen or by growers themselves (94, 103). InOregon, mixtures are commonly sold, even by the largest companies (103), andmixtures provide an attractive market niche for some seed producers (C.C. Mundt,unpublished). Some apprehension has been expressed regarding shifts of mixturecomponents over time for small grain cultivar mixtures in some environments(167). In other localities, wheat cultivar mixtures are carried over for 3 to 4 yearswith few problems regarding component shifts (103).

Though mixture components must be compatible in terms of agronomic andquality characteristics, the difficulty of attaining such compatibility has often beenoveremphasized (36, 103, 167). For example, 13% and 18% of the Oregon (131)and Washington (164) winter wheat areas are currently being grown as cultivarmixtures, with no apparent agronomic or marketing difficulties. Further, there of-ten is as much difference among plants or tillers of the same cultivar grown inthe same field as there is between different cultivars in a mixture. Considerableheight differences can often be accommodated by harvest machinery; occasion-ally, height differences even facilitate harvest (103). Height differences betweencultivars that are too large, however, may necessitate processing more straw orstalk material than is desired (15, 103) and may require excessive attention by theoperator to ensure that all of the crop is harvested. The importance of variationin crop maturity date will depend on the crop species and geographical locationin question. In the Pacific Northwest region, where conditions become increas-ingly hot and dry as the wheat crop matures, large maturity differences at anthesistranslate into very small difference at harvest. Thus, there is little problem in mix-ing wheat cultivars with different maturity (103), except at the highest elevationswhere cooler temperatures sometimes extend maturity differences. Compatibilityfor maturity date of field crop cultivars may be more critical in areas where summerrainfall is common, however [see (15)], and may be even more critical for somevegetable and fruit crops.

Differences in quality and marketing of mixtures will depend on the crop inquestion. For many field crops, the harvested product is often not segregated bycultivar upon sale, and environment can often have a greater influence on qualitycharacteristics than does cultivar (103). Given that cultivar mixtures produce wheat


400 MUNDT

flour (36, 56, 103), wheat grain (145), barley malt (123, 155), barley grain (58),and rice grain (25) with quality equal to or better than the mean of its componentcultivars in pure stand, there should usually be no difficulty involved with saleof mixtures containing genotypes that produce acceptable quality when grownalone. Indeed, mixtures can sometimes be used to improve the quality of a product(36, 56, 145). For crops that are sold to the consumer by cultivar (e.g., apples),it would be necessary to harvest the cultivars separately, sort by cultivar afterharvest, or to develop multiline cultivars consisting of components with highlysimilar quality characteristics.

It is now uncommon for plant breeders to produce true pure-line cultivars ofmany field crops, and within-cultivar diversity may improve yield stability and re-duce vulnerability to disease. For example, the most successful groundnut (Arachishypogea) cultivars grown in the United States (e.g., ‘Florigiant’ and ‘Florerunner’)are mixtures developed by bulking phenotypically similar lines in the F3 to F5 gen-erations (70). These cultivars have long dominated U.S. groundnut production, andhave demonstrated a high level of stability over environments for yield and othercharacteristics (69). Interestingly, the soft white winter wheat cultivar Stephenswas developed in a similar manner, but through the accidental bulking of linesearly in the breeding process. Released in 1977, Stephens has been one of the mostsuccessful wheat cultivars in the United States, showing tremendous productivityand adaptability, and still comprised half of the Oregon winter wheat area in 2001(131). Though largely anecdotal, such observations suggest that retaining diversitywithin cultivars may be highly beneficial in traditional breeding programs, and thatdiversity should be introduced into cultivars produced via biotechnology, for exam-ple, by transforming resistance genes into multiple host genetic backgrounds andby bulking phenotypically similar progeny resulting from a doubled haploid cross.

CONCLUSIONS

It is now 50 years since the pioneering call for use of multiline crop cultivarsby Jensen (61). The usefulness of mixtures for control of powdery mildews andrusts of small grains, which account for the majority of mixture research, has beendemonstrated repeatedly through experimental studies. There have been somemajor successes in using mixtures to control such diseases at a commercial scale.However, mixtures will be more useful under some epidemiological conditionsthan under others, and experimental methodology, especially scale issues, may becrucial in evaluating the potential usefulness of mixtures.

A data base is now accumulating to evaluate the effect of mixtures for othertypes of pathosystems. In the rare cases when all other variables can be heldconstant experimentally, and in mathematical models, it is possible to demonstratethat mixtures are, on average, less effective under some conditions, e.g., for cropswith large plants. All else is rarely equal in the real world, however, and there areexamples of mixtures providing both low and high degrees of disease control for



almost any type of pathosystem imaginable, including crops with large plants, andpathogens that demonstrate low host specificity, or are splash dispersed, soilborne,or insect vectored.

Despite the absence of strong evidence for rapid loss of mixture effective-ness owing to selection of complex pathogen genotypes, our knowledge in thisarea is inadequate. Empirical studies to address this issue are difficult to conduct,and mathematical models have taken a simplistic view by focusing mostly onstatic costs of virulence as the main mechanism countering selection for pathogencomplexity in mixtures. Many other mechanisms that may potentially maintainpathogen diversity in mixtures need to be investigated. Studies of pathogen evolu-tion need to consider both relative and absolute frequencies of pathogen genotypes,and mixtures should be managed to avoid strong selection for complex pathogengenotypes that are highly fit.

Theory and models have been useful in evaluating epidemic development andpathogen evolution in mixtures, especially when combined with results of empiri-cal studies. However, current models do not describe events in the field accurately,and results from well-designed field experiments continue to limit our understand-ing of the effect of mixtures on disease development. Owing to host and pathogencomplexity, field studies of mixtures are more difficult to conduct and more proneto error and bias than are many other types of field studies.

In the past, many agricultural researchers have assumed that mixtures will notprovide adequate disease control, are impractical, or would prove to be unaccept-able to farmers. These arguments are clearly invalid as generalities, and thereare now many examples of mixtures contributing substantially to disease control,yield increases, and yield stability in commercial agriculture. On the other hand,the concept of functional diversity instructs that not all mixtures will provide ben-efits, and that use of mixtures will not be feasible in all cases. Multilines, cultivarmixtures, and other methods of genetic diversification should be considered asimportant tools for management of all crops, but must be appropriately evaluatedwithin the context of the needs of the local agricultural industry and through directinteraction with farmers. Mixtures will not be the disease control tactic of choicein all cases. Given the need for a more sustainable agriculture based on models ofnatural ecosystems, however, host mixtures will likely play a much larger role inthe next 50 years than they have in the past half century.

The Annual Review of Phytopathologyis online at http://phyto.annualreviews.org

LITERATURE CITED

1. Ahmed HU, Finckh MR, Alfonso AF,Mundt CC. 1997. Epidemiological effectof gene deployment strategies on bacte-rial blight of rice.Phytopathology87:66–70

2. Akanda SI, Mundt CC. 1996. Effects oftwo-component wheat cultivar mixtureson stripe rust severity.Phytopathology86:347–53

3. Alexander HM, Roelfs AP, Cobbs G.


402 MUNDT

1986. Effects of disease and plant compe-tition on yield in monocultures and mix-tures of two wheat cultivars.Plant Pathol.35:457–65

4. Allan RE, Line RF, Peterson CJ Jr, Ruben-thaler GL, Morrison KJ, Rohde CR. 1983.Crew, a multiline wheat cultivar.Crop Sci.23:1015–16

5. Assefa H, Van den Bosch F, Zadoks JC.1995. Focus expansion of bean rust in cul-tivar mixtures.Plant Pathol.44:503–9

6. Ayanru DKG, Browning JA. 1977. Ef-fect of heterogeneous oat populations onthe epiphytotic development of victoriablight. New Phytol.79:613–23

7. Barrera JAS, Frederiksen RA. 1994. Eval-uation of sorghum hybrid mixtures forcontrolling sorghum leaf-blight.PlantDis. 78:499–503

8. Barrett JA. 1980. Pathogen evolution inmultilines and variety mixtures.Z. Pfl-anzenkr. Pflanzenschutz.87:383–96

9. Barrett JA. 1981. Disease progress curvesand dispersal gradients in multilines.Phy-topathol. Z.100:361–96

10. Barrett JA, Wolfe MS. 1980. Pathogen re-sponse to host resistance and its impli-cations in breeding programmes.Eur.Mediterr. Plant Prot. Organ. Bull.10:341–47

11. Berger RD, Bergamin A, Amorim L.1997. Lesion expansion as an epidemiccomponent.Phytopathology87:1005–13

12. Blaise P, Gessler C. 1994. Cultivar mix-tures in apple orchards as a means to con-trol apple scab? InIntegrated Control ofPome Fruit Diseases, ed. DJ Butt, pp.105–12.Norw. J. Agric. Sci. Suppl.17

13. Bonman JM, Khush GS, Nelson RJ. 1992.Breeding rice for resistance to pests.Annu. Rev. Phytopathol.30:507–28

14. Boudreau MA, Mundt CC. 1997. Ecologi-cal approaches to disease control. InEnvi-ronmentally Safe Approaches to Crop Dis-ease Control, ed. J Rechcigl, N Rechcigl,pp. 33–62. Boca Raton, FL: CRC Press

15. Bowden R, Shroyer J, Roozeboom K,Claassen M, Evans P, et al. 2001.Perfor-

mance of Wheat Variety Blends in Kan-sas. Kansas State Univ. Agric. Ext. Bull.128 (www.oznet.ksu.edu/library/crpsl2/SRL128.pdf)

16. Brophy LS, Mundt CC. 1991. Influence ofplant spatial patterns on disease dynamics,grain yield and plant competition in genet-ically diverse wheat populations.Agric.Ecosyst. Environ.35:1–12

17. Browning JA, Frey KJ. 1969. Multilinecultivars as a means of disease control.Annu. Rev. Phytopathol.14:355–82

18. Browning JA, Frey KJ. 1981. The multi-line concept in theory and practice. SeeRef. 60a, pp. 37–46

19. Buiel AAM, Verhaar MA, van den BoschF, Hoogkamer W, Zadoks JC. 1989. Ef-fect of cultivar mixtures on the wave ve-locity of expanding yellow rust foci inwinter wheat.Neth. J. Agric. Sci.37:75–78

20. Burdon JJ, Chilvers GA. 1977. Control-led environment experiments on epidemicrates of barley mildew in different mix-tures of barley and wheat.Oecologia28:141–46

21. Burdon JJ, Chilvers GA. 1982. Host den-sity as a factor in disease ecology.Annu.Rev. Phytopathol.20:143–66

22. Calonnec A, Goyeau H, de Vallavieille-Pope C. 1996. Effects of induced resis-tance on infection efficiency and sporula-tion of Puccinia striiformison seedlingsin varietal mixtures and on field epidemicsin pure stands.Eur. J. Plant Pathol.5:733–41

23. Castro A. 2001.Cultivar Mixtures. ThePlant Health Instructor, DOI: 101094/PHI-A-2001-1230-01. St. Paul: Am.Phytopathol. Soc. (www.apsnet.org/edu-cation/AdvancedPlantPath/top.html)

24. Chakraborty S, Pettitt AN, Cameron DF,Irwin JAG, Davis RD. 1991. Anthracnosedevelopment in pure and mixed stands ofthe pasture legumeStylosanthes-scabra.Phytopathology81:788–93

25. Chin KM, Husin AN. 1982. Rice vari-ety mixtures. InProc. Padi Workshop,



Bumbong Lima, Province Wellesley, Mal-aysia, pp. 203–16

26. Chin KM, Wolfe MS. 1984. Selectionon Erysiphe graminisin pure and mixedstands of barley.Plant Pathol.33:535–46

27. Chin KM, Wolfe MS. 1984. The spreadof Erysiphe graminisf. sp.hordeiin mix-tures of barley varieties.Plant Pathol.33:89–100

28. Cohan FM, King EC, Zawadzki P.1994. Amelioration of the deleteriouspleiotropic effects of an adaptive mutationin Bacillus subtilis. Evolution48:81–95

29. Cournoyer BM. 1967.Crown rust inten-sification within and dissemination frompureline and multiline oat varieties. PhDthesis. Iowa State Univ., Ames

29a. Cowger C, Mundt CC. 2002. Effects ofwheat cultivar mixtures on epidemic pro-gression of Septoria tritici blotch andpathogenicity ofMycosphaerella gram-inicola. Phytopathology92: In press

29b. Csada RD, James PC, Espie RHM.1996. The “filedrawer problem” of non-significant results: Does it apply to bio-logical research?Oikos76:591–93

30. DiLeone JA, Mundt CC. 1994. Effect ofwheat cultivar mixtures on populations ofPuccinia striiformisraces.Plant Pathol.43:917–30

31. Dinoor A, Eshed N. 1984. The role andimportance of pathogens in natural plantcommunities. Annu. Rev. Phytopathol.22:443–66

32. Dubin HJ, Wolfe MS. 1994. Compara-tive behavior of three wheat cultivars andtheir mixture in India, Nepal, and Pak-istan.Field Crops Res.39:71–83

33. Elliot VJ, MacKenzie DR, Nelson RR.1980. Effect of number of componentlines on powdery mildew epidemics in awheat multiline.Phytopathology70:461–62 (Abstr.)

34. Ferrandino FJ. 1993. Dispersive epidemicwaves: I. Focus expansion within a linearplanting.Phytopathology83:795–802

35. Finckh MR, Gacek ES, Czembor HJ,Wolfe MS. 1999. Host frequency and den-

sity effects on disease and yield in mix-tures of barley.Plant Pathol.48:807–16

36. Finckh MR, Gacek ES, Goyeau H, Lan-nou C, Merz U, et al. 2000. Cereal va-riety and species mixtures in practice.Agronomie20:813–37

37. Finckh MR, Mundt CC. 1992. Plant com-petition and disease in genetically diversewheat populations.Oecologia91:81–92

38. Finckh MR, Mundt CC. 1992. Stripe rust,yield, and plant competition in wheat cul-tivar mixtures.Phytopathology82:905–13

39. Finckh MR, Wolfe MS. 1997. The useof biodiversity to restrict plant diseasesand some consequences for farmers andsociety. In Ecology in Agriculture, ed.LE Jackson, pp. 199–233. San Diego:Academic

40. Finckh MR, Wolfe MS. 1998. Diversifi-cation strategies. InThe Epidemiology ofPlant Diseases, ed. DG Jones, pp. 231–59.London: Chapman & Hall

41. Fitt BDL, McCartney HA. 1986. Sporedispersal in relation to epidemic models.In Plant Disease Epidemiology, Vol. 1, ed.KJ Leonard, WE Fry, pp. 311–45. NewYork: MacMillan

42. Fitt BDL, McCartney HA, Walklate PJ.1989. The role of rain in dispersal of path-ogen inoculum.Annu. Rev. Phytopathol.27:241–70

43. Fleming RA. 1983. Development of asimple mechanistic model of cereal rustprogress.Phytopathology73:308–12

44. Francis CA, ed. 1986.Multiple CroppingSystems. New York: MacMillan

45. Garrett KA, Mundt CC. 1999. Epidemi-ology in mixed host populations.Phy-topathology89:984–90

46. Garrett KA, Mundt CC. 2000. Effects ofplanting density and the composition ofwheat cultivar mixtures on stripe rust:an analysis taking into account limits tothe replication of controls.Phytopathol-ogy90:1313–21

47. Garrett KA, Mundt CC. 2000. Host diver-sity can reduce potato late blight severity


404 MUNDT

for focal and general patterns of primaryinoculum.Phytopathology90:1307–12

48. Garrett KA, Nelson RJ, Mundt CC, Cha-con G, Jaramillo RE, Forbes GA. 2001.The effects of host diversity and othermanagement components on epidemics ofpotato late blight in the humid highlandtropics.Phytopathology91:993–1000

49. Gieffers W, Hesselbach J. 1988. Diseaseincidence and yield of different cerealcultivars in pure stands and mixtures. I.Spring barley (Hordeum vulgareL.). Z.Pflanzenkrankh. Pflanzenschutz95:46–62

50. Gieffers W, Hesselbach J. 1988. Diseaseincidence and yield of different cerealcultivars in pure stands and mixtures. II.Winter barley (Hordeum vulgareL.). Z.Pflanzenkrankh. Pflanzenschutz95:63–69

51. Goleniewski G, Newton AC. 1994. Mod-elling the spread of fungal diseases using anearest neighbor approach.Plant Pathol.43:631–43

52. Gumpert FM. 1989. Measuring diseaseprogress in pure and mixed stands of plantcultivars.Phytopathology79:968–73

53. Halloin JM, Johnson DJ. 2000. Reduc-tion of sugarbeet losses from Rhizocto-nia crown and root rot by use of mixturesof resistant and susceptible varieties.Phy-topathology90:S33 (Abstr.)

54. Hariri D, Fouchard M, Prud’homme H.2001. Incidence ofSoil-borne wheat mo-saic virusin mixtures of susceptible andresistant wheat cultivars.Eur. J. PlantPathol.107:625–31

55. Huang R, Kranz J, Welz HG. 1994. Selec-tion of pathotypes ofErysiphe graminisf. sp.hordei in pure and mixed stands ofspring barley.Plant Pathol.43:458–70

56. Jackson LF, Wennig RW. 1997. Use ofwheat cultivar blends to improve grainyield and quality and reduce disease andlodging.Field Crops Res.52:261–69

57. James WC, Shih CS, Callbeck LC, Hodg-son WA. 1973. Interplot interference infield experiments with late blight of potato(Phytophthora infestans). Phytopathol-ogy63:1269–75

58. Jedel PE, Helm JH, Burnett PA. 1998.Yield, quality and stress tolerance of bar-ley mixtures in central Alberta.Can. J.Plant Sci.78:429–36

59. Jeger MJ, Griffiths E, Jones DG. 1981.Disease progress of non-specialised fun-gal pathogens in intraspecific mixedstands of cereal cultivars. I. Models.Ann.Appl. Biol.98:187–98

60. Jeger MJ, Jones DG, Griffiths E. 1981.Disease progress of non-specialised fun-gal pathogens in intraspecific mixedstands of cereal cultivars. II. Field exper-iments.Ann. Appl. Biol.98:199–210

60a. Jenkyn JF, Plumb RT, eds. 1981.Strate-gies for the Control of Cereal Disease.Ox-ford: Blackwell

61. Jensen NF. 1952. Intra-varietal diversifi-cation in oat breeding.Agron. J.44:50–54

62. Kardin MK, Groth JV. 1989. Density-dependent fitness interactions in the beanrust fungus.Phytopathology79:409–12

63. Karjalainen R. 1986. Spring wheat mix-tures in northern crop production: abilityof mixtures to buffer disease developmentand yield loss caused bySeptoria nodo-rum. J. Agric. Sci. Finland58:33–42

64. Karjalainen R, Peltonen-Sainio P. 1993.Effect of oat cultivar mixtures on di-sease progress and yield reductioncaused by barley yellow dwarf virus.Z.Pflanzenkrankh. Pflanzenschutz100:58–68

65. Katsuya K, Green GJ. 1967. Reproductivepotentials of races 15B and 56 of wheatstem rust.Can. J. Bot.45:1077–91

66. Kessler K. 1997. Wheat blends stabilizeyields.Furrow, Jan., pp. 27–28

67. Kiyosawa S. 1989. Breakdown of blast re-sistance in relation to general strategiesof resistance gene deployment to prolongeffectiveness of resistance in plants. InPlant Disease Epidemiology, Vol. 2, ed.KJ Leonard, WE Fry, pp. 251–83. NewYork: McGraw-Hill

68. Kiyosawa S, Shiyomi M. 1972. A theo-retical evaluation of the effect of mixingresistant variety with susceptible variety



for controlling plant diseases.Ann. Phy-topathol. Soc. Jpn.38:41–51

69. Knauft DA, Gorbet DW. 1991. Agro-nomic performance and genetic shifts ofgenotype mixtures in peanut.Euphytica52:85–90

70. Knauft DA, Norden AJ, Gorbet DW. 1987.Peanut. InPrinciples of Cultivar Devel-opment, ed. WR Fehr, pp. 346–84. NewYork: MacMillan

71. Knott EA, Mundt CC. 1990. Combiningability analysis of wheat cultivar mixturesunder diseased and non-diseased condi-tions.Theor. Appl. Genet.80:313–20

72. Koizumi S. 2001. Rice blast control withmultilines in Japan. See Ref. 94a, pp. 143–57

73. Koizumi S, Kato H. 1987. Effect of mixedplantings of susceptible and resistant ricecultivars on leaf blast development.Ann.Phytopathol. Soc. Jpn.53:28–38

74. Kolmer JA. 1995. Selection ofPucciniarecondita f.sp. tritici virulence pheno-types in three multilines of Thatcherwheat lines near isogenic for leaf rust re-sistance genes.Can. J. Bot.73:1081–88

75. Kousik CS, Sanders DC, Ritchie DF.1996. Mixed genotypes combined withcopper sprays to manage bacterial spot ofbell peppers.Phytopathology86:502–8

76. Lannou C. 2001. Intra-pathotype diversityfor aggressiveness and pathogen evolu-tion in cultivar mixtures.Phytopathology91:500–10

77. Lannou C, de Vallavieille-Pope C, Bi-ass C, Goyeau H. 1994. The efficacyof mixtures of susceptible and resistanthosts to two wheat rusts with differentlesion size: controlled condition experi-ments and computer simulations.J. Phy-topathol.140:227–37

78. Lannou C, de Vallavieille-Pope C, Goy-eau H. 1994. Host mixture efficacy indisease control: effects of lesion growthanalyzed through computer-simulatedepidemics.Plant Pathol.43:651–62

79. Lannou C, de Vallavieille-Pope C, Goy-eau H. 1995. Induced resistance in host

mixtures and its effect on disease controlin computer-simulated epidemics.PlantPathol.44:478–89

80. Lannou C, Mundt CC. 1996. Evolution ofa pathogen population in host mixtures.I. Study of the simple race-complex raceequilibrium.Plant Pathol.45:440–53

81. Lannou C, Mundt CC. 1997. Evolutionof a pathogen population in host mix-tures: rate of emergence of complex races.Theor. Appl. Genet.94:991–99

82. Leach JE, Vera Cruz CM, Bai J, Leung H.2001. Pathogen fitness penalty as a pre-dictor of durability of disease resistancegenes.Annu. Rev. Phytopathol.39:187–224

83. Leonard KJ. 1969. Factors affecting ratesof stem rust increase in mixed plantingsof susceptible and resistant oat varieties.Phytopathology59:1845–50

84. Leonard KJ, Czochor RJ. 1980. Theoryof genetic interactions among populationsof plants and their pathogens.Annu. Rev.Phytopathol.18:237–58

85. Levin BR, Veronique P, Walker N. 2000.Compensatory mutations, antibiotic re-sistance and the population genetics ofadaptive evolution in bacteria.Genetics154:985–97

86. Lopez CG, Mundt CC. 2000. Using mix-ing ability analysis from two-way culti-var mixtures to predict the performanceof cultivars in complex mixtures.FieldCrops Res.68:121–32

87. Luthra JK, Rao MV. 1979. Escape mech-anism operating in multilines and its sig-nificance in relation to leaf rust epidemics.Indian J. Genet. Plant Breed.39:38–49

88. Luthra JK, Rao MV. 1979. Multilinecultivars—how their resistance influencesleaf rust epidemics in wheat.Euphytica28:137–44

89. MacHardy WE, Gadoury DM, GesslerC. 2001. Parasitic and biological fitnessof Venturia inequalis: relationship to dis-ease management strategies.Plant Dis.85:1036–51


406 MUNDT

90. Madden LV. 1992. Rainfall and the dis-persal of fungal spores.Adv. Plant Pathol.8:39–79

91. Marshall DR. 1989. Modeling the effectsof multiline varieties on the population ge-netics of plant pathogens. InPlant DiseaseEpidemiology, Vol. 2, ed. KJ Leonard, WEFry, pp. 284–317. New York: McGraw-Hill

92. McCracken AR, Dawson WM. 1998.Short rotation coppice willow in North-ern Ireland since 1973: development ofthe use of mixtures in the control of fo-liar rust (Melampsoraspp.).Eur. J. For.Pathol.28:241–50

93. McDonald BA, Allard RW, Webster RK.1988. Responses of two-, three-, and four-component barley mixtures to a variablepathogen population.Crop Sci.28:447–52

94. Merz U, Valenghi D. 1997. ‘Extenso’ pro-duction and cereal mixtures in Switzer-land. In Variety Mixtures in Theoryand Practice, ed. MS Wolfe, pp. 9–10. Grignon, France: COST Action 817(www.scri.sari.ac.uk/TiPP)

94a. Mew TW, Borromeo E, Hardy B, eds.2001. Exploiting Biodiversity for Sus-tainable Pest Management, Proc. ImpactSymp. Exploit. Biodivers. Sustain. PestManag., Kunming, China, 21–23, Aug.2000. Los Ba˜nos, Philippines: Int. RiceRes. Inst.

95. Milgroom MG. 1990. A stochastic modelfor the initial occurrence and developmentof fungicide resistance in plant pathogenpopulations.Phytopathology80:410–16

96. Minogue KP, Fry WE. 1983. Models forthe spread of disease: model description.Phytopathology73:1168–73

97. Miot S, Frey P, Pinon J. 1999. Varietalmixture of poplar clones: effect on in-fection by Melampsora larici-populinaand on plant growth.Eur. J. For. Pathol.29:411–23

98. Moreno-Ruiz G, Castillo-Zapata J. 1990.The variety Colombia: a variety of coffeewith resistance to rust (Hemileia vastatrix

Berk & Br). Cent. Nac. Invest. Cafe, Ceni-cafe, Chinchina-Caldas, Colombia. Tech.Bull. 9

99. Morrell V. 1997. Antibiotic resistance:road of no return.Science278:575–76

100. Muller K, McDermott JM, Wolfe MS,Limpert E. 1996. Analysis of diversityin populations of plant pathogens: thebarley powdery mildew pathogen acrossEurope.Eur. J. Plant Pathol.102:385–95

101. Mundt CC. 1990. Disease dynamics inagroecosystems. InAgroecology, ed. CRCarroll, JH Vandermeer, PM Rossett, pp.263–69. New York: McGraw-Hill

102. Mundt CC. 1994. Techniques to managepathogen co-evolution with host plantsto prolong resistance. InRice Pest Sci-ence and Management, ed. PS Teng,KL Heong, K Moody, pp. 193–205. LosBanos, Philippines: Int. Rice Res. Inst.

103. Mundt CC. 1994. Use of host genetic di-versity to control cereal diseases: impli-cations for rice blast. InRice Blast Dis-ease, ed. S Leong, RS Zeigler, PS Teng,pp. 293–307. Cambridge: CABI Int.

104. Mundt CC. 1997. Disease increase in hostmixtures. InExercises in Plant DiseaseEpidemiology, ed. L Francl, D Nehr, pp.119–22. St. Paul: APS Press

105. Mundt CC. 2001. Varietal diversificationand disease control. See Ref. 94a, pp. 135–42

106. Mundt CC. 2002. Performance of wheatcultivars and cultivar mixtures in thepresence ofCephalosporiumstripe.CropProt. In press

107. Mundt CC, Brophy LS. 1988. Influenceof number of host genotype units on theeffectiveness of host mixtures for dis-ease control: a modeling approach.Phy-topathology78:1087–94

108. Mundt CC, Brophy LS, Kolar SC. 1996.Effect of genotype unit number and spa-tial arrangement on severity of yellow rustin wheat cultivar mixtures.Plant Pathol.45:215–22

109. Mundt CC, Brophy LS, Schmitt ME.



1995. Choosing crop cultivars and mix-tures under high versus low disease pres-sure: a case study with wheat.Crop Prot.14:509–15

110. Mundt CC, Brophy LS, Schmitt ME.1995. Disease severity and yield of pure-line wheat cultivars and mixtures in thepresence of eyespot, yellow rust, andtheir combination.Plant Pathol.44:173–82

111. Mundt CC, Browning JA. 1985. Devel-opment of crown rust epidemics in ge-netically diverse oat populations: effectof genotype unit area.Phytopathology75:607–10

112. Mundt CC, Browning JA. 1985. Geneticdiversity and cereal rust management. InThe Cereal Rusts, Vol. 2, ed. AP Roelfs,WR Bushnell. Orlando: Academic

113. Mundt CC, Cowger C, Garrett KA. 2002.Relevance of integrated management toresistance durability.Euphytica.In press

114. Mundt CC, Cowger C, Hoffer ME. 1999.Disease management using variety mix-tures. InSeptoria and Stagonospora Dis-eases of Cereals: A Compilation of GlobalResearch, ed. M van Ginkel, A McNab,J Krupinsky, pp. 111–16. Mexico, DF:CIMMYT

115. Mundt CC, Hayes PM, Sch¨on CC. 1994.Influence of variety mixtures on severityof scald and net blotch and on the yield ofbarley.Plant Pathol.43:356–61

116. Mundt CC, Karow RS. 1995. How doesdisease affect club wheat performance?Oregon Wheat, July, pp. 12–13

117. Mundt CC, Leonard KJ. 1985. Effect ofhost genotype unit area on epidemic de-velopment of crown rust following focaland general inoculations of mixtures ofimmune and susceptible oat plants.Phy-topathology75:1141–45

118. Mundt CC, Leonard KJ. 1986. Analysisof factors affecting disease increase andspread in mixtures of immune and sus-ceptible plants in computer-simulated epi-demics.Phytopathology76:832–40

119. Mundt CC, Leonard KJ. 1986. Effect of

host genotype unit area on development offocal epidemics of bean rust and commonmaize rust in mixtures of resistant and sus-ceptible plants.Phytopathology76:895–900

120. Mundt CC, Leonard KJ, Thal WM, Ful-ton JH. 1986. Computerized simulationof crown rust epidemics in mixtures ofimmune and susceptible oat plants withdifferent genotype unit areas and spa-tial distributions of initial disease.Phy-topathology86:590–98

121. Munk L. 1983. Response of a powderymildew population to a barley variety mix-ture. InDurable Resistance in Crops, ed.F Lamberti, JM Waller, NA Van der Graaf,pp. 105–7. New York: Plenum

122. Newton AC, Ellis RP, Hackett CA, GuyDC. 1997. The effect of component num-ber onRhynchosporium secalisinfectionand yield in mixtures of winter barley cul-tivars.Plant Pathol.45:930–38

123. Newton AC, Swanston JS, Guy DC, EllisRP. 1998. The effect of cultivar mixtureson malting quality in winter wheat.J. Ind.Brew.104:41–51

124. Newton AC, Thomas WTB. 1992. The ef-fect of specific and non-specific resistancein mixtures of barley genotypes on infec-tion by mildew (Erysiphe graminisf. sp.hordei) and on yield.Euphytica59:73–81

125. Newton AC, Thomas WTB. 1993. The in-teraction of either an effective or a de-feated major gene with non-specific re-sistance on mildew infection (Erysiphegraminis f. sp. hordei) and yield in mix-tures of barley.J. Phytopathol.139:268–74

126. Newton MR, Kinkel LL, Leonard KJ.1997. Competition and density-dependentfitness in a plant parasitic fungus.Ecology78:1774–84

127. Newton MR, Kinkel LL, Leonard KJ.1998. Determinants of density- and fre-quency-dependent fitness in competingplant pathogens.Phytopathology88:45–51


408 MUNDT

128. Ntahimpera N, Dillard HR, Cobb AC,Seem RC. 1996. Colletotrichum develop-ment in mixtures of resistant and suscep-tible dry bean cultivars.Phytopathology86:668–73

129. Nugugi HK, King SB, Holt J, Julian AM.2001. Simultaneous temporal progress ofsorghum anthracnose and leaf blight incrop mixtures with disparate patterns.Phytopathology91:720–29

130. Ohara RB, Brown JKM. 1996. Frequency-and density-dependent selection in wheatpowdery mildew.Heredity77:439–47

131. Oregon Agric. Stat. Serv. 2000.1999–2000 Oregon Agriculture & FisheriesStatistics. Salem, OR: Oregon Dep. Agric.

132. Ostergaard H. 1983. Predicting develop-ment of epidemics on cultivar mixtures.Phytopathology73:166–72

133. Parlevliet JE. 1981. Stabilizing selectionin crop patho-systems: an empty conceptor a reality?Euphytica30:259–69

134. Parlevliet JE, van Ommeren A. 1975.Partial resistance of barley to leaf rust,Puccinia hordei. II. Relationship betweenfield trials, micro plot tests and latent pe-riod. Euphytica24:293–303

135. Paysour RE, Fry WE. 1983. Interplot in-terference: a model for planning fieldexperiments with aerially disseminatedpathogens.Phytopathology73:1014–20

136. Pfahler PL, Linskens HF. 1979. Yieldstability and population diversity in oats(Avenasp.).Theor. Appl. Genet.54:1–5

137. Pfleeger TG, Mundt CC. 1998. Wheat leafrust severity as affected by plant densityand species proportion in simple com-munities of wheat and wild oats.Phy-topathology88:708–14

138. Pielou EC. 1977.Mathematical Ecology.New York: Wiley

139. Power AG. 1987. Plant community diver-sity, herbivore movement, and an insect-transmitted disease of maize.Ecology68:1658–69

140. Power AG. 1990. Cropping systems, in-sect movement, and the spread of insect-transmitted diseases in crops. InAgroecol-

ogy, ed. SR Gliessman, pp. 47–69. NewYork: Springer-Verlag

141. Power AG. 1991. Virus spread and vec-tor dynamics in genetically diverse plantpopulations.Ecology72:232–41

142. Priestly RH. 1981. Choice and deploy-ment of resistant cultivars for cereal rustcontrol. See Ref. 60a, pp. 65–72

143. Priestly RH, Bayles R. 1982. Varietaldiversification as a means of reducingthe spread of cereal diseases in theUnited Kingdom.J. Natl. Inst. Agric. Bot.15:205–14

144. Robinson RA. 1976.Plant Pathosystems.New York: Springer-Verlag

145. Sarandon SJ, Sarandon R. 1995. Mixtureof cultivars: pilot field trial of an ecolog-ical alternative to improve production orquality of wheat (Triticum aestivum). J.Appl. Ecol.32:288–94

146. Saur L, Mille B. 1997. Disease progressof Pseudocercosporella herpotrichoidesin mixed stands of winter wheat cultivars.Agronomie17:113–18

147. Schmidt RA. 1978. Diseases in forestecosystems: the importance of functionaldiversity. InPlant Disease: An AdvancedTreatise, Vol. 2, ed. JG Horsfall, EB Cowl-ing, pp. 287–315. New York: Academic

148. Segal A, Browning JA, Fishbeck G, WahlI. 1982. Balance in indigenous plant pop-ulations. InResistance to Diseases andPests in Forest Trees, ed. HM Heybroek,BR Stephan, K von Weissenberg, pp. 361–70. Wageningen: PUDOC

149. Segal A, Manisterski J, Fishbeck G,Wahl I. 1980. How plant populations de-fend themselves in natural ecosystems. InPlant Disease: An Advanced Treatise, Vol.5, ed. JG Horsfall, EB Cowling, pp. 75–102. New York: Academic

150. Sharma RC, Dubin HJ. 1996. Effect ofwheat cultivar mixtures on spot blotch(Bipolaris sorokiniana). Field Crops Res.48:95–101

151. Shaw MW, Ostergard H. 1998. Furthercomments on models of pathogen co-evolution in cultivar mixtures and the



definition of pathogenic fitness.Phy-topathology88:492–93

152. Sitch L, Whittington WJ. 1983. The effectof variety mixtures on the developmentof swede powdery mildew.Plant Pathol.32:41–46

153. Smithson JB, Lenne JM. 1996. Varietalmixtures: a viable strategy for sustain-able productivity in subsistence agricul-ture.Ann. Appl. Biol.128:127–58

154. Sun P, Yang XB. 1999. Two properties ofa gene-for-gene coevolution system un-der human perturbations.Phytopathology89:811–16

155. Swanston JS, Newton AC, Guy DC,Gacek ES. 2000. Malting performance ofbarley cultivar mixtures from the UK andPoland.J. Ind. Brew.106:239–43

156. Tapsoda H, Wilson JP. 1999. Increasingcomplexity of resistance in host popula-tions through intermating to manage rustof pearl millet.Phytopathology89:450–55

157. Thurston HD. 1992.Sustainable Prac-tices for Plant Disease Management inTraditional Farming Systems. Boulder,CO: Westview

158. Van den Bosch F, Frinking F, Metz JAJ,Zadoks JC. 1990. Focus expansion inplant disease. IV. Expansion rates in mix-tures of resistant and susceptible hosts.Phytopathology80:598–602

159. van den Bosch F, Zadoks JC, Metz JAJ.1988. Focus expansion in plant disease.I: The constant rate of focus expansion.Phytopathology78:54–58

160. van den Bosch F, Zadoks JC, Metz JAJ.1988. Focus expansion in plant disease. II.Realistic parameter-sparse models.Phy-topathology78:59–64

161. Vanderplank JE. 1963.Plant Diseases:Epidemics and Control. New York: Aca-demic

162. Vera Cruz CM, Bai J, Ona I, Leung H,Nelson RJ, et al. 2000. Predicting durabil-ity of a disease resistance gene based onan assessment of the fitness loss and epi-demiological consequences of avirulence

gene mutation.Proc. Natl. Acad. Sci. USA97:13500–5

163. Villareal LMMA, Lannou C. 2000. Se-lection for increased spore efficacy byhost genetic background in a wheat pow-dery mildew population.Phytopathology90:1300–6

164. Washington Agric. Stat. Serv. 2000.Washington Agricultural Statistics 2000.Olympia, WA: Wash. State Dep. Agric.

165. Wilcox JR, St. Martin SK. 1998. Soybeangenotypes resistant toPhytophthora so-jae and compensation for yield losses ofsusceptible isolines.Plant Dis. 82:303–6

166. Wilson JP, Gates RN, Panwar MS. 2001.Dynamic multiline population approachto resistance gene management.Phy-topathology91:255–60

167. Wolfe MS. 1985. The current status andprospects of multiline cultivars and vari-ety mixtures for disease resistance.Annu.Rev. Phytopathol.23:251–73

168. Wolfe MS. 1992. Maintaining the value ofour varieties. InBarley Genetics VI, Vol.2, ed. L Munk, pp. 1055–67. Copenhagen:Munksgaard Int. Publ.

169. Wolfe MS, Barrett JA. 1980. Can we leadthe pathogen astray?Plant Dis.64:148–55

170. Wolfe MS, Barrett JA, Jenkins JEE. 1981.The use of cultivar mixtures for diseasecontrol. See Ref. 60a, pp. 73–80

171. Wolfe MS, Brandle U, Koller B, LimpertE, McDermott JM, et al. 1992. Barleymildew in Europe: population biology andhost resistance.Euphytica63:125–39

172. Wolfe MS, McDermott JM. 1994. Pop-ulation genetics of plant pathogen in-teractions: the example of theErysiphegraminis-Hordeum vulgarepathosystem.Annu. Rev. Phytopathol.32:89–113

173. Wolfe MS, Schwarzbach E. 1978. Pat-terns of race changes in powdery mildews.Annu. Rev. Phytopathol.16:159–80

174. Xu X-M, Ridout MS. 2000. Stochasticsimulation of the spread of race-speci-fic and race-nonspecific aerial fungal


410 MUNDT

pathogens in cultivar mixtures.PlantPathol.49:207–18

175. Yang XB, Ferrandino FJ. 1998. Linearmodels to estimate relative parasitic fit-ness: a reply.Phytopathology88:490–91

176. Yong L, Zadoks JC. 1992. A decisionmodel for variety mixtures to control yel-low rust on winter wheat.Agric. Syst.38:17–33

177. Zeigler RS, Cuoc LX, Scott RP, BernardoMA, Chen DH, et al. 1995. The rela-tionship between lineage and virulencein Pyricularia oryzaein the Philippines.Phytopathology85:443–51

178. Zhu Y, Chen H, Fan J, Wang Y, Li Y,et al. 2000. Genetic diversity and dis-ease control in rice.Nature 406:718–22

Documents

C. C. Mundt - plantpath.cornell.eduplantpath.cornell.edu/Courses/pp601/MGM lectures 2004/Mundt 2002... · 5 Jul 2002 10:25 AR AR165-PY40-14.tex AR165-PY40-14.SGM LaTeX2e(2002/01/18)